REPLICATION FILES FOR:

Crossley, T.F., and H. Low, �Job Loss, Credit Constraints and Consumption Growth.� Review of Economics and Statistics, 96(5):876-884 (December, 2014) 


CONTACT: tfcrossley@gmail.com or tcross@esex.ac.uk


**********************************************************************************************************************************************
1. PROGRAMS

This package contains the following stata do-files

1. cl_data.do cleans up the survey data and prepares a file for analysis
3. the following do-files which are called by cl_data.do
	skip95.do
	roevars.do
	dates95.do
	educ95.do
	cl_avariables;
	cl_sample.do
   they should be placed in a subfolder called �subroutines�
2. cl_credit.do does the analysis. 

********************************************************************************************************************************************
2. DATA

The Public Use Microdata File (PUMF) for the 1995 CANADIAN OUT OF EMPLOYMENT PANEL is described here:

http://www23.statcan.gc.ca/imdb/p2SV.pl?Function=getSurvey&SurvId=1148&InstaId=29813&SDDS=4423

Each user must obtain their own copy from STATISTICS CANADA, but everyone is eligible to do so, and will be charged just the cost of a CD. 

The data can be ordered by emailing STATISTICS CANADA at: infostats@statcan.gc.ca 
(If you have difficult at this step we are happy to try and help).

The estimates in the published paper are based on a pre-release version of the data. We have subsequently learned that the PUMF differs in two ways.
a. Statistics Canada has deleted any respondent who attritted between the first and second interview of the panel from the PUMF. This leads to smaller samples sizes and slightly different numbers for some of the descriptive statistics. Note however, that it does not affect our key consumption growth estimates in Tables 4 and 5 (as these are based on households responding to both waves of the panel). 
b. The variable that captures household income at job loss (*rhhinc*) involved some imputation in the data we used and this has changed. Again this does not affect our key tables (4 and 5) but does result in minor differences in some of the prior descriptive statistics. 

We include in this package a data extract (cl2014.dta) which matches the PUMF data in these respects and which can be run directly by cl_credit.do and a log file for comparison with the published estimates. 


**********************************************************************************************************************************************




